There is a paper which adds to the discussion of long term tropospheric water vapor trends. It is
Shannon Brown, Shailen Desai, Stephen Keihm, Wenwen Lu and Christopher Ruf; 2007 Ocean water vapor and cloud burden trends derived from the Topex Microwave Radiomoeter.
The abstract reads
“An end-of-mission recalibration effort was recently completed for Topex Microwave Radiometer to generate climate data records of precipitable water vapor and cloud liquid water for 1992-2005. The TMR climate data is analysed for trends. The global trend in precipitable water vapor is found to be 0.9 + 0.06 mm/decade. Regional precipitable water vapor trends are found to be highly correlated with regional sea surface temperature trends. The cloud liquid water trends are observed to be generally negative outside the tropics and positive in the tropics.”
The association of sea surface temperature (SST) trend with tropospheric water vapor is concluded, as summarized in Figures 6 and 7, but, unfortunately, the explained variance is only 21% for all SST trends and 13% for SSTs greater than 15 C. Nonetheless, there is a physical reason that larger amounts of tropospheric water vapor should be expected (due to great evaporation from the warmer ocean surface).
There is a more substantive issue, however, with the trend values that they present. While the authors focus on the linear trends from 1992 to 2005 which they plot in each of the panels in Figure 2, the trends since about 2002 have been flat, or in the case of the 0-60N values, have even been negative!
This lack of continued moistening of the troposphere since about 2002 is consistent with the conclusions of Climate Science that
1. The 60S-60N SSTs have cooled since 2002 (see)
2. The water vapor content over land has not been increasing (see).
3. The added information that results when we examine shorter time periods (particularly the most recent years) should be encouraged for all papers that examine trends and variability in time of climate metrics, including tropospheric water vapor content.